Archive - Edition 2: Reporting

back To Overview of Edition

Leading Article

Current Trends in Reporting

Content

Information Visualization

Selected Portals

 

Creating Reporting Structures Through Information Modeling

by Dr. JoAnn T. Hackos, President Comtech Services, Inc. http://www.comtech-serv.com – 12/22/2000

Disclaimer: Please note that this edition was written in 2000. Therefore, statements in the articles, particularly those regarding SAP's products, product strategy, branding strategy, and organizational structure, may no longer be valid.

Abstract

Much of the activity involved in improving user interface designs has focused on improving methods of entering information. Our interface designs are concerned with making the "input" part of the interface as intuitive as possible for users to provide information to databases and programming algorithms. Yet, equally important to our user community is the information we provide back to them. They view reports and other data that enable them to proceed with actions or make critical decisions before acting. A key in designing an effective Information Model that meets the decision-making needs of the users is understanding the complex and often shifitng conceptual models of the audience, that is the recognition that hierarchical or linear structures for reporting information are often ineffective.

Over the last two years, I have been working with the concept of Information Modeling as a way to frame information structures so that information becomes more accessible to our users and supports their decision-making. In Information Modeling, we examine the assumptions that users make about information structures and how they assume information elements are supposed to relate to one another. We look for evidence that suggests that information grouped in particular ways will be most easily interpreted and used for further analysis and decision making.

I initially became interested in Information Modeling about 15 years ago when I worked on an operations research project with the Mathematics Department at the University of Colorado. We were creating decision-support software that contained enormously complex algorithms based on an artificial-intelligence design. The software was designed to assist complex decision making. Unfortunately, the processes were so complex that users tended to distrust the results. They simply could not understand why a particular decision was made by the system. Consequently, they tended to ignore the system and make their own "gut" decisions.

We were tasked with figuring out a way to report the results of the software’s calculations in an understandable manner. What resulted was an experimental back-end tool that used graphic representations to partially chart the processes and display some of the interrelationships among the data that led to the decisions. This graphic back-end tool remained experimental but the user research taught us a lot about how people viewed reporting structures.

At the same time, I was made more aware than ever of the deficiencies of data reporting and the struggle that we put users through when we present them with raw data. The basic message users get from us is "Here’s all the information. Good luck in finding what you need."

For example, we have been working recently with optical network systems. One of the most important reports from these systems is the list of alarms. Users note current alarms in the system by accessing an alarm list, presented in alphabetical order. They need to find the most severe alarms or the alarms most likely to invoke still other alarms, locate alarms that are linked by cause-and-effect, and then clear the alarms in a sequence from the most severe to the least. Often, clearing a more severe alarm results in other less severe alarms actually disappearing from the system. The process is quite complex and involves the need for a fairly sophisticated understanding of how the alarm system functions.

Despite the complexity of the task, the interface gives little structure to the alarm reports to aid in the alarm clearing process. Users are pretty much on their own.

 

Understanding users’ conceptual information models

To know how to present information effectively to our users, we must concentrate our investigations on understanding existing conceptual models, especially in terms of the context in which information is understood. We need a comprehensive understanding of what users need to know, when they need to know it, and in what ways they attempt to interpret information and turn it into analysis and action.

For example, we have spent considerable time working on an Information Model of patient information, first to serve the needs of physicians and other health workers in a general clinical setting and, most recently, to enhance the diagnosis and treatment of cancer patients.

In the first case, we were trying to understand how physicians use a complex network of information about a patient’s condition to diagnose problems and find solutions. We learned, for example, that information must be understood over time, a linear information structure. A patient’s blood pressure viewed in an instant of time tells less about that patient’s overall health than blood pressures viewed and understood over a long period of time. A linear structure, perhaps best illustrated with a time-dependent graph, provides an informed view of this data.

At the same time, we learned that one piece of data is insufficient because many pieces of data are interrelated by the physician to form a decision. Information on blood pressure must be viewed in relationship to information on medications, since many medications, including even popular over-the-counter medications, influence blood pressure. In addition, physicians take into account general information about a patient’s overall physical condition over time and look for changes that spell trouble.

When trying to construct a tool to present patient information and assist with diagnoses, we learned that information must not only be viewed over time but also holistically. When we presented a tabbed interface in which we had grouped certain types of information together, we learned that the physicians lost the ability to view the multiple points of information that enhance diagnoses.

As a result of user testing and workplace observations, we learned that user configurability is a key to designing successful reporting structures. If we could let the physicians rearrange the information to suit their mental model of the task, they would be better able to understand its significance and act accordingly. We also needed to provide flexibility in rearranging at a moment’s notice because our physicians often needed to change the relationships among the information to help them understand an immediate problem. Their focus shifted as certain information increased in significance while other information became secondary.

The need for a constantly shifting focus of attention in this complex user environment has also led us to a significant new understanding. That is the recognition that hierarchical or linear structures for reporting information are often ineffective. Instead of hierarchy or time line, we have moved now, in many instances, to a hub and spoke arrangement to solve a number of information-reporting problems.

The hub of the information model represents the user’s most immediate focus of attention – whatever key information takes precedence at a particular moment in the analysis or interrelationships. The spokes represent related information that may be viewed, discarded, viewed again, and perhaps moved to the hub position as the interrelationships change.

With the advent of the portal concept for Internet delivery, however, we once again have the possibility of examining non-hierarchical methods of providing structure to information. The hub-and-spokes model appears promising as long as we can assist the user in identifying relationships among data.

Note that the key issue in designing an effective interface to report data is to understand exactly how users relate to information. Sometimes a hierarchical relationship will be most effective, but a hierarchy requires that the user already has considerable knowledge of the information structure. A tabbed interface, for example, represents a set hierarchy of information elements. As long as users know what they are likely to find behind a tab and as long as the information on each tab is reasonably independent of other information, this interface can be effective. Time-dependent relationships are excellent for information in which changes over time are significant. However, time-dependent data may be difficult to relate to other non-time-dependent data. A hub-and-spokes Information Model builds on the interrelationships recorded from analyses of expert users and observations of the way current users relate to the data. They imply flexibility to the extent that they can be dynamically restructured based on active user intervention or continuing observations of behavior.

In a recent article in Fast Company on the Web site, weather.com, the information architect explained that they have chosen a similar hub-and-spoke architecture for their information. The new architecture enables users to access related information based on their current focus much more quickly than if they were to navigate a hierarchy of information.

Of course, the major requirement in developing a network of interrelated information that has meaning to the user is to know the user’s conceptual model of the information being delivered. With that conceptual model in place, we have the possibility of creating an Information Model that is effective in reporting data and related information so that it meets the decision-making needs of the audience.


Biography

Dr. JoAnn Hackos is President of Comtech, an user-centered-design consulting firm based in Denver, which she founded in 1978; Director of the Center for Information-Development Management, a member-sponsored organization for information development and training management issues; and co-founder and partner of SingleSource Associates. Dr. Hackos is called upon by major corporations to consult on the management of their interface development, information models, and the information design, including GUI interfaces, Web-based information, and documentation databases. She and her staff members are also involved in research studies involving the assessment of customer needs related to product usability, technical information, and training.

For more than 20 years, Dr. Hackos has conducted seminars internationally on subjects ranging from project management, designing effective interfaces and information, minimal information products, usability testing, online documentation and computer-based training, to managing the information-design and development process. The seminars are dedicated to enhancing the practices and products that will best promote customer satisfaction.

Dr. Hackos is co-author of User and Task Analysis for Interface Design with Ginny Redish (Wiley 1998). Her book Managing Documentation Projects (Wiley, 1994) is widely regarded as the bible of publications management. Standards for Online Communication (Wiley 1997) is a compendium of industry process and publications standard practices. JoAnn is a Fellow and Past President of the international Society for Technical Communication (STC), recipient of the Arthur Goldsmith award of the IEEE, and honorary member of the Conseil de Redacteurs.

Recent clients include Varian Medical Systems, US West, Dell Computer, Cadence Design Systems, SAP, Alcatel, Ericsson, Nokia Networks, Motorola, Nortel Networks, Federal Express, Lucent Technologies, Compaq Computer, and more.

 

top top